RTplace: A Reinforcement Learning-based Macro Placement Method with ResNet and Transformer

As the scale and complexity of integrated circuit design continue to expand, automated chip placement technology has become crucial for enhancing design efficiency, reducing costs, and optimizing chip performance. This paper proposes a reinforcement learning-based macro placement method, which combines ResNet with Transformer, which autonomously generates high-quality chip placements through a global and local feature fusion approach(RTplace). RTplace embeds Transformer into ResNet structure to obtain different types of features. Additionally, a multi-level feature fusion module is introduced to achieve the fusion of global and local features. The pro-posed method captures information that might be lost during feature propagation and enhances the quality of placement. The experimental results indicate that RTplace has better placement performance compared to other existing methods.